AI Frontier Institute

News & Research

The latest AI research and news with real-world stakes. Each item is sourced, dated, summarized in plain English and tagged by impact area, and checked against its source before it appears.

• ResearchNature Health2026-04-16WEQP

  Public use of a generalist LLM chatbot for health queries · Beatriz Costa-Gomes, Pavel Tolmachev, Eloise Taysom et al.

  This study analyzes over 500,000 de-identified health-related conversations with Microsoft Copilot to characterize how people use generalist AI chatbots for health queries. Nearly one in five conversations involves personal symptom assessment or condition discussion, and one in seven personal health queries concerns someone other than the user, suggesting conversational AI serves a caregiving role. Personal health queries spike in the evening and nighttime when traditional healthcare is least accessible, and usage patterns diverge sharply by device. The findings have direct implications for platform-specific design, safety considerations, and the responsible development of health AI.
• ResearchInternational Journal of Environmental Research and Public Health2026-04-16WEQP

  Healthcare Providers’ Perceptions and Multi-Level Determinants of Adoption of an AI-Powered Electrocardiography Interpretation Clinical Decision Support System in Ethiopia: A Formative Qualitative Study · Minyahil Tadesse Boltena, Ziad El-Khatib, Amare Zewdie et al.

  This qualitative study from Ethiopia examined how healthcare providers perceive and would adopt an AI-powered ECG interpretation clinical decision support system (CDSS) across ten hospitals in four regions. Through 31 in-depth interviews with cardiologists, nurses, general practitioners, and critical care specialists, researchers identified six key themes including perceived benefits, trust, workflow integration, and ethical concerns. Providers saw the AI tool as a valuable adjunct—not a replacement—for clinicians, capable of improving diagnostic accuracy, reducing subjectivity, and cutting unnecessary referrals, but stressed that high accuracy, validation, training, and leadership support are prerequisites for adoption. The findings highlight that context-sensitive implementation strategies and multi-level system readiness are essential for deploying AI-based diagnostic tools in low-resource healthcare settings.
• ResearchFrontiers in Artificial Intelligence2026-04-16WEQP

  A structured framework for effective and responsible generative artificial intelligence chatbot prompt engineering throughout the scientific process: a comprehensive guide for the health and medical researcher · Jeremy Y. Ng

  This paper presents a 10-chapter structured framework guiding health and medical researchers in using generative AI chatbots responsibly throughout the full research cycle, from question development and study design to data analysis and dissemination. The authors emphasize prompt engineering—crafting clear, purposeful prompts—as a key skill for improving AI output quality across varied methodological contexts. The guide addresses risks including hallucinated content, embedded biases, and ethical challenges, stressing that AI-generated content must be verified and that human expertise, critical thinking, and methodological rigor remain irreplaceable. It concludes that GenAI tools should augment, not replace, researcher judgment, and that transparency and accountability in AI use are essential.
• ResearchJournal of theoretical and applied electronic commerce research2026-04-16EQP

  Effect of Explainable AI Features on User Satisfaction and Purchase Intention in Saudi Mobile Shopping Apps · Ahmed S. M. Almamy, Sufyan Habib, Layla K. Nasser et al.

  This study surveyed 597 mobile shoppers in Saudi Arabia to assess how explainable AI (XAI) features—fairness, trustworthiness, transparency, and interpretability—affect satisfaction and purchase intention in e-commerce apps. Using PLS-SEM within a stimulus–organism–response framework, the researchers found that fairness and bias detection, trustworthiness, and transparency significantly shape consumers' cognitive and affective states, which in turn drive satisfaction and purchase intention, with consumer satisfaction acting as a key mediator. Notably, interpretability had limited impact, suggesting users prioritize fairness and trust over technical explanations. The findings offer actionable guidance for mobile shopping platforms to design AI systems that reduce perceived bias and enhance transparency to improve consumer engagement and competitive advantage.
• ResearchRemote Sensing2026-04-16EQP

  Artificial Intelligence and Machine Learning in Remote Sensing for Tropical Forest Monitoring: Applications, Challenges, and Emerging Solutions · Belachew Gizachew

  This review examines how AI and machine learning integrated with remote sensing are transforming tropical forest monitoring under frameworks like REDD+ and the Paris Agreement's Enhanced Transparency Framework. The authors find that combining optical, radar, and LiDAR time series within cloud-based platforms has substantially improved automation, scalability, and speed of deforestation and biomass/carbon monitoring. However, adoption remains constrained by limited training data, reliance on proprietary platforms, uneven institutional capacity, and unresolved governance challenges. The review identifies open datasets, platform-agnostic infrastructure, and inclusive data-governance frameworks as critical enablers for nationally owned monitoring systems.
• ResearchClinical Neurophysiology2026-04-16QCP

  IFCN position statement: use of artificial intelligence in clinical neurophysiology · Aatif M. Husain, Sandor Beniczky, Meriem Bensalem-Owen et al.

  The International Federation of Clinical Neurophysiology (IFCN) has issued a position statement outlining guidelines for the responsible integration of AI in clinical neurophysiology (CNP), covering modalities such as EEG and EMG. The statement emphasizes that AI should serve as a decision-support tool rather than a replacement for clinical expertise, and that models deployed without expert supervision must meet stringent standards for accuracy, safety, and quality control. Key requirements include dataset diversity, transparent and ethical training practices, continuous performance monitoring, and clinician feedback to maintain system reliability. The IFCN envisions AI advancing CNP through real-time decision support, personalized interventions, and remote monitoring, while prioritizing patient safety and clinical accountability.
• ResearchSustainability2026-04-16WEP

  Artificial Intelligence in the Labor Market: Evidence on Worker Inclusion, Exclusion, and Discrimination—A Systematic Review · Carlos Rouco, Paula Figueiredo, Carlos Pedro Gonçalves et al.

  This systematic review synthesizes peer-reviewed evidence on how AI applications in labor markets—including recruitment, performance management, and algorithmic work management—affect worker inclusion, exclusion, and discrimination. Analyzing 19 eligible studies via qualitative thematic synthesis (PRISMA 2020), the review finds an ambivalent pattern: AI can support inclusion through assistive technologies and better matching, but can also worsen occupational polarization, digital exclusion, and discriminatory outcomes when models are trained on biased data or deployed without transparency. The authors propose a hybrid governance model integrating technical audits, organizational audits, inclusive upskilling, participatory regulation, and responsible HR policies across four interdependent layers to align AI innovation with decent and inclusive work.
• ResearchComputer law & security review2026-04-16EQP

  AI, climate, and regulation: From data centers to the AI Act · Kai Ebert, Nicolas Alder, Ralf Herbrich et al.

  This paper examines the intersection of AI regulation and climate change, focusing on the EU AI Act and data center regulations. The authors find that the AI Act fails to address energy consumption from AI inferences and indirect greenhouse gas emissions from AI applications, and they propose a specific interpretation to bring these gaps into regulatory scope. The paper makes twelve concrete policy proposals across four areas—energy and environmental reporting, legal clarifications, transparency and accountability, and future far-reaching measures—including recommending measurement at the cumulative server level and incorporating sustainability risk assessments into mandatory AI Act risk evaluations. The findings matter for policymakers seeking to align AI governance with climate accountability obligations.
• ResearchZenodo (CERN European Organization for Nuclear Research)2026-04-16ECP

  The Competitive Deployment Dilemma: Market Incentives, Collective Action, and the Policy Conditions for Accountable AI Authority · Alexander Huseby

  This paper argues that the persistent gap between AI deployment and accountability frameworks is a collective action problem: individual organizations rationally underinvest in AI governance because costs are private and immediate while benefits—such as systemic risk avoidance and public institutional trust—are shared and diffuse, creating free-rider dynamics. The authors examine three mechanisms to close this gap: mandatory minimum standards via regulation, market mechanisms that differentially price governance quality, and polycentric coordination for shared governance infrastructure. Drawing on the EU AI Act, AI liability insurance markets, ISO/IEC 42001 certification, and the limitations of voluntary commitments like the Seoul Frontier AI Safety Commitments (2024), the paper identifies conditions under which governance investment becomes a self-sustaining competitive advantage rather than a burden. The findings matter because they reframe AI governance failure as a structural market problem requiring coordinated policy interventions rather than simply better organizational ethics.
• ResearchInternational Journal of Law and Social Sciences2026-04-16WEP

  Artificial Intelligence and the Future of Human Rights: Legal Accountability for Algorithmic Decision-Making in Democratic Societies. · Bagus Khusfi Satyo, S.H., M.Kn., Bagus Agung Yuda Prasetyo, S.H., M.Kn

  This study examines the legal accountability of algorithmic decision-making systems in democratic societies, finding that traditional legal frameworks are often insufficient to address the complex challenges posed by AI in areas such as employment, healthcare, law enforcement, and public administration. Using qualitative doctrinal and comparative legal methodology, the research identifies risks including algorithmic bias, opacity, privacy violations, and weakened procedural accountability. The authors conclude that effective AI governance requires new regulatory approaches emphasizing transparency, explainability, and human oversight, and that a rights-based regulatory framework is essential to balance innovation with the protection of fundamental freedoms. The study calls for stronger institutional oversight, clearer liability allocation among governments and technology companies, and international cooperation on AI governance.
• ResearchZenodo (CERN European Organization for Nuclear Research)2026-04-16ECP

  The Competitive Deployment Dilemma: Market Incentives, Collective Action, and the Policy Conditions for Accountable AI Authority · Alexander Huseby

  This paper argues that the gap between expanding AI decision-making authority and accountability frameworks persists because it is a collective action problem: individual organizations rationally underinvest in AI governance since costs are private and immediate while benefits such as avoided systemic risk and public trust are shared and diffuse, creating free-rider incentives. The authors examine three structural mechanisms that could close this gap—mandatory minimum standards via regulation, market mechanisms that price governance quality differentially, and polycentric coordination for shared governance infrastructure—drawing on the EU AI Act, AI liability insurance markets, ISO/IEC 42001 certification, and the limitations of voluntary commitments like the Seoul Frontier AI Safety Commitments. The paper concludes by specifying conditions under which governance investment becomes self-sustaining as a competitive advantage rather than a competitive burden. The findings have direct implications for how regulation, certification markets, and industry coordination can incentivize accountable AI deployment.
• ResearchPhilPapers (PhilPapers Foundation)QP

  Regulating the Frontier: Model-Level Evidence from the EU AI Act · Han Li

  This study provides an early empirical assessment of whether the EU AI Act's transparency obligations for general-purpose AI models are changing how developers document their models in practice. Using a database of 3,571 machine-learning models and a seven-component Structured Disclosure Index, the authors find an 8.7 percentage-point increase in structured disclosure among general-purpose language models after the Act's obligations took effect, concentrated in training-compute reporting, though the estimate loses statistical significance in more stringent specifications. The findings suggest selective documentation adaptation rather than substantive behavioral change, and the authors distinguish between 'structured transparency' (filing disclosures) and 'substantive transparency' (actual openness in model development). The paper contributes a reproducible measurement framework relevant to evaluating AI regulation effectiveness.
• ResearchDigital USD (University of San Diego)EP

  Algorithmic Oversight: Caremark's Fiduciary Framework Applied to Artificial Intelligence · Samar Singh

  This paper argues that corporate boards have a fiduciary duty under Delaware's Caremark doctrine—as refined by Marchand v. Barnhill, In re Boeing, and In re McDonald's—to actively oversee AI systems that drive core business functions such as loan approvals and hiring. It identifies a structural 'black box problem' where AI systems can produce unlawful outcomes while generating clean compliance reports, meaning traditional red-flag oversight mechanisms are insufficient. The paper proposes a scalable governance framework including board-level AI audit committees, a Chief AI Officer with reporting duties, mandatory governance disclosure, and a safe harbor for compliance. It concludes that existing Delaware law already provides the doctrinal tools needed to govern AI, and urges boards to act proactively before litigation compels them to do so.
• ResearchKnowledge Commons (Lakehead University)P

  Algorithmic Justice and Responsible AI Journalism: A Comparative Communication Policy Perspective in East Asia · Renxiudi Huang, Linjia Bai

  This paper compares how China and South Korea regulate AI-driven journalism and algorithmic content curation, examining the tension between algorithmic fairness and platform accountability. China's top-down state-centric model emphasizes ideological security, algorithmic registration, and synthetic media labeling, while South Korea's co-regulation model relies on data protection, anti-monopoly intervention, and civil society oversight. The study analyzes key policy documents and platforms (Toutiao and Naver), finding that both approaches reduce risks of algorithmic bias and disinformation but face contrasting trade-offs between regulatory efficiency and editorial independence. The authors propose a 'Policy-Media-Society' (PMS) governance framework to advance transparent, verified public information access aligned with UN SDG 16.

← Back to the dashboard